Statistical and mathematical modeling to evaluate the cost-effectiveness of Helicobacter pylori screening and treating strategies in Mexico in the setting of antibiotic resistance

Abstract

Helicobacter pylori (H. pylori), a bacterium that is present in the stomach of half of the world’s population with disproportionate burden in developing countries, is the strongest known biological risk factor for gastric cancer. Gastric cancer is the fourth most common type of cancer and the second cause of cancer death in the world. In particular, in Mexico gastric cancer is the third highest cause of cancer death in adults, with some regions having cancer mortality rates that are twice the national average (8.0 vs. 3.9 per 100,000, respectively). H. pylori can be treated with antibiotics, but widespread treatment may lead to significant levels of antibiotic resistance (ABR). ABR is one of the main causes of H. pylori treatment failure and represents one of the greatest emerging global health threats. In this thesis, we use statistical and mathematical modeling to investigate the health benefits, harms, costs and cost-effectiveness of screen-and-treat strategies for identifying and treating persons with H. pylori to inform public health practice in three steps. First, we estimated the age-specific force of infection of H. pylori --defined as the instantaneous per capita rate at which susceptibles acquire infection-- using a novel hierarchical nonlinear Bayesian catalytic epidemic model with data from a national H. pylori seroepidemiology survey in Mexico. Second, we developed an age-structured, susceptible-infected-susceptible (SIS) transmission model of H. pylori infection in Mexico that included both treatment-sensitive and treatment-resistant strains. Model parameters were derived from the published literature and estimated from primary data. Using the model, we projected H. pylori infection and resistance levels over 20 years without treatment and for three hypothetical population-wide treatment policies assumed to be implemented in 2018. In sensitivity analyses, we considered different mixing patterns and trends of background antibiotic use. We validated the model against historical values of prevalence of infection and ABR of H. pylori. Third, we expanded the SIS model to incorporate the natural history of gastric carcinogenesis including gastritis, intestinal metaplasia, dysplasia and ultimately non-cardia gastric cancer. We then estimated the cost-effectiveness of various screen-and-treat strategies for H. pylori infection and ABR in the Mexican population from the health sector perspective.